Evaluation of CI In-Cylinder Flow using optical and numerical techniques 1 1 2 2 R. Rezaei , S. Pischinger , P. Adomeit , J. Ewald * 1 Institute for Combustion Engines, RWTH Aachen University, Germany 2 FEV Motorentechnik GmbH, Aachen, Germany In order to evaluate different port concepts for modern Compression-Ignition engines, usually quantities as the swirl level and the flow coefficient are evaluated, which are measured on a stationary flow test bench. As additional crite- rion, in this work, the homogeneity of the swirl flow is introduced and defined quantitatively. Different valve lift strate- gies are evaluated using three-dimensional Particle Imaging Velocimetry in a stationary flow configuration and tran- sient In-Cylinder CFD simulation using both the Reynolds Averaged Navier Stokes equation and the Large Eddy simulation approach. Introduction New concepts for High-Speed Direct Injection • Basic engine: FEV system engine 205 Compression Ignition (CI) engines are under de- • Bore x stroke: 75 mm x 88.3 mm velopment due to the increased awareness of the • Injection 200 system: BOSCH 8.0 2000 bar Piezo ISFC [g/kWh] mm maximum Valve Lift CO2 emission impact on global climate change • Compression 195 ratio: 15.3 6.4 mm maximum Valve Lift 4.8 mm maximum Valve Lift which goes hand in hand with the demand of fur- 3.2 mm maximum Valve Lift ther reduced fuel consumption as well as the ag- As is190 shown in Figure 1, a reduction of the valve gravated emission legislation standards. lift provides 185 the best potential for emission beha- In order to meet theses requirements for future vior by increasing the swirl ratio. The utilization of 180 CI engines, not only the injection system has to be the increased homogeneous swirl by reducing the 2.5 205 suitably defined. Also, for the chosen injection valve lift reduces smoke emission significantly [g/kWh] [-] system, the optimal in-bowl swirl has to be gener- without200 any impact on fuel consumption. 2.0 A further ISFC Number 8.0 mm maximum Valve Lift 6.4 mm maximum Valve Lift ated. The magnitude of the swirl optimum, howev- reduction 1.5 195 of valve lift leads to a noteworthy in- 4.8 mm maximum Valve Lift er, is furthermore dependent on the operating point crease of gas exchange losses, which finallyValve 3.2 mm maximum leadsLift and engine speed. Therefore, in order to provide to increased 1.0 190 fuel consumption without any advan- Smoke the corresponding flexibility, a CI engine concept tage concerning 0.5 185 soot emission. has been developed that features a variable valve 0.0 180 lift and port deactivation concept. By means of this,205 [bar] 2.5 0.0 the optimal trade-off between swirl level and in- Number [-] 200 ISFC [g/kWh] cylinder fresh charge filling level can be found. 2.0 -0.1 8.0 mm maximum Valve Lift Gas Exchange 6.4 mm maximum Valve Lift It has been shown that different valve lift strate-195 1.5 -0.2 4.8 mm maximum Valve Lift gies nominally lead to similar filling and swirl le- 3.2 mm maximum Valve Lift vels. However, differences in combustion behavior190 1.0 -0.3 IMEPSmoke and engine-out emissions give rise to the assump- 0.5 -0.4 tion that local differences in the in-cylinder flow185 -0.5 0.0 structure caused by different valve lift strategies180 IMEP Gas Exchange [bar] 0.0 0.5 1.0 1.5 2.0 0.0 have noticeable impact. 2.5 NOx-Emission [g/kWh] In this work, these flow structures were ana- -0.1 Smoke Number [-] lyzed and quantitatively assessed using both opti- 2.0 Figure 1: 1500 rpm, 6.8 bar emissions [6] -0.2 cal and numerical techniques. 1.5 -0.3 Increasing the swirl via reduced valve lift pro- Engine: Variable Charge Motion concept 1.0vides -0.4a slight improvement to the particulate The intake port of this DI diesel engine consists air/fuel ratio trade-off from 8.0 mm to 3.2 mm valve -0.5 of tangential and filling ports [1]. Tangential ports 0.5lift. In Figure 0.0 2 in 0.5 particular1.0the NO1.5 x/soot trade-off 2.0 can be used to generate a relatively high swirl ratio is shown in particular for two valve lift strategies, while the filling ports, as the name already implies, 0.0maximum lift of 4.8NOmm x -Emission [g/kWh] vs. 8 mm max. valve lift IMEP Gas Exchange [bar] provide a high flow coefficient. Additionally the 0.0with the filling port closed, which have the same intake charge flow can be directed by machining swirl ratio. It was seen that the soot emissions with the valve seat rings to yield swirl chamfers. This-0.1a closed filling port are considerably higher than for concept enables the generation of extremely high-0.2a lift of 4.8 mm. Therefore, the swirl level alone is swirl numbers with low valve lifts without reducing insufficient to describe the in-cylinder flow. This the flow for high valve lifts. The impact of different-0.3was also found in [4]. valve strategies on the combustion system using a single-cylinder engine is assessed [1]. The test-0.4 engine had the following design parameter: -0.5 0.0 0.5 1.0 1.5 2.0 * Corresponding author: ewald_j@fev.de NOx-Emission [g/kWh] Towards Clean Diesel Engines, TCDE 2009 and also the measured swirl ratio ( CU C A ) at the PIV test bench. CU C A RMSVtheta RMSVtheta CU C A Filling port deac- 2.28 3.94 1.73 tivated 3.2 mm Both ports 5.66 3.79 0.67 active 1.6 mm Table 1: Results of swirl ratio and RMS of the tan- gential velocity. As can be seen, the in-cylinder flow field gener- ated when both ports are active is more homoge- neous than the case with port deactivation. Figure 2: 2280 rpm 9.4 bar, emissions [1] Computational Setup PIV measurements of stationary intake port In this study, the commercial CFD software flow STAR-CD is used for the transient calculations of 3D PIV stationary flow analysis of the new port the intake and compression stroke with moving design was performed for various valve lifts, and valves and piston. On the intake and exhaust port port deactivation strategies. flange positions, pressure boundary conditions both ports active - 1.6 mm valve lift from GT-Power gas exchange calculations were tangential Port employed. The calculation were performed from filling Port 360°CA to 720°CA. Two different turbulence mod- els, the LES Smagorinsky [2] and also the k-ε model [3] are used for intake flow simulations. Characterization of In-Cylinder Flow in- homogeneity Swirl Velocity Distribution RMS Velocity Fluctuation In order to quantify the in-homogeneity of the filling port deactivated - 3.2 mm valve lift in-cylinder flow field different cut sections perpen- dicular to the cylinder axis are considered. Each of the considered cut sections is divided into concen- tric rings, shown in Figure 4. 1 23 45 6 Swirl Velocity Distribution RMS Velocity Fluctuation Figure 3: Charge motion analysis by 3D PIV without and with filling port deactivation at z = 75 mm Figure 3 shows the flow field in a horizontal Figure 4: Top view of a cut section considering six section 75 mm below the cylinder head. The aver- rings age flow distribution is displayed as a vector field on the left of Figure 3 and the local distribution of For each ring, first a mean value of the tangen- the flow fluctuation intensity is displayed by the tial velocity component is calculated. Then, for velocity RMS on the right side of the same figure. each of these rings the RMS of the tangential ve- In both cases the intake flow rates are similar, but locity is determined. the resulting swirl flow patterns are strongly differ- ent. For the filling port deactivation, the swirl flow Simulation results using RANS and LES structure is less coherent, and fluctuation intensity The in-cylinder angular velocity is defined as is increased. angular momentum divided by the moment of iner- Table 1 compares the Root Mean Square tia. (RMS) value of the measured tangential velocity – - (1) 4.8 mm valve lift 4.8 mm valve lift 0 Filling port closed Filling port closed 0 Distance from cylinder head [mm] Distance from cylinder head [mm] 2 The dimensionless swirl ratio for each operating 4 2 4 6 point is then obtained according to 8 6 8 10 10 12 12 in  cylinder 14 14 Swirl ratio  . (2) 16 16 Engine 18 18 20 20 22 22 0 1 2 3 4 5 6 7 8 9 10 0 1 2 3 4 5 6 7 8 9 10 Results of CFD simulations using the different RMSVtheta [m/s] RMSVtheta [m/s] valve lift and port strategies are shown in Figure 5. Figure 7: RMS value of tangential velocity using RANS(left graph) and LES(right graph) It can be seen that the in-cylinder swirl ratio can be increased to the same level either by reducing the As can be seen, differences in in-cylinder flow maximum valve lift to 4.8mm or by port deactiva- field between these valve strategies can be distin- tion. guished using the LES turbulence model rather 9 8.0 mm maximum valve lift than the k-ε model. The investigations with the 8 6.4 mm maximum valve lift 4.8 mm maximum valve lift LES model show that the 4.8 mm valve lift produc- 7 3.2 mm maximum valve lift Filling port closed es more homogeneous swirl than port deactivation. 6 Swirl ratio [-] This is also in agreement with experimental inves- 5 4 tigations at the 3D PIV flow test. 3 2 Summary and conclusions 1 Differences in emission behavior for different 0 valve lifts and with and without deactivated filling 360 420 480 540 600 660 720 Crank angle [°CA ATDC] port were observed for a single-cylinder engine. Measurements on a stationary flow bench and Figure 5: Swirl ratio over crank angle as calculated CFD calculation both assessed the swirl level for by RANS CFD simulations. the different concepts. While in particular a maxi- mum valve lift of 4.8 mm for both intake ports pro- Inhomogeneity of in-cylinder flow duces the same swirl level as the maximum valve Figure 6 compares the cut sections of the tan- lift of 8 mm with the filling port deactivated, the gential velocity field of each strategy at the middle soot emissions are significantly different and high- of the piston bowl which is simulated using the k-ε er for the latter configuration model in STAR-CD at 2280 rpm. Therefore it was argued that next to the swirl ra- Using RANS model tio, another important parameter, describing these Using LES model Tangential velocity Tangential velocity discrepancies, is required. An approach to eva- [m/s] [m/s] luate the in-homogeneity of in-cylinder flow by Filling port Filling port closed closed means of PIV and CFD simulation was developed and presented. While in CFD, the RANS approach could not show visible differences in the in- homogeneity of in-cylinder flow, the LES approach valve lift valve lift 4.8 mm 4.8 mm in CFD and the 3D-PIV method showed differenc- es between two cases in a way that the in- homogeneity in the case filling port closed is higher Figure 6: Cut section of the tangential velocity field in than a maximum valve lift of 4.8 mm for both the middle of the bowl using the RANS model (left) and valves. the LES model (right) at -30°CA ATDC This work presented here has been submitted As from the RANS simulation, there is almost to the SAE ICE conference in September 2009. no difference between the two cases in terms of References predicting the in-homogeneity of the flow field. [1] D. Adolph, R. Rezaei, S. Pischinger, P. Adomeit, T. A cut section of the tangential velocity field in Körfer, A. Kolbeck, M. Lamping, D. Tatur, D. Tomaz- the middle of the bowl is shown also in Figure 6 for ic, Gas exchange optimization and the effect on the same valve lift strategies using the LES model. Emission reduction for HSDI diesel engines, SAE The LES turbulence model captures turbulent flow Paper 2009-01-0653 (2009). [2] J. Smagorinsky, General Circulation experiments structures while in RANS only the mean flow is with the primitive equations I. The basic experiment, resolved. Monthly Weather Review, 91(3),99-164 (1963). The RMS values of the tangential velocity from [3] B.E. Launder, D.B. Spalding. The numerical compu- RANS and LES are compared for both valve strat- tation of turbulent flow. Computer Methods in Applied egies in Figure 7. Mechanics and Engineering 3,269-289 (1972) [4] P. Adomeit, S. Pischinger, M. Becker, H. Rohs, A. Greis, G. Grünefeld, Potential Soot and CO Reduc- tion for HSDI Diesel Combustion Systems, SAE Pa- per 2006-01-1417 (2006)